change filter_stream to filter

[grosser]

I wrote some other plugins and alway used filter, so wanted to check which is best and https://docs.fluentd.org/filter#filter-chain-optimization says to not use filter_stream ... is there any reason to keep it or can it be changed as long as the tests pass ?

[jcantrill]

I wrote some other plugins and alway used filter, so wanted to check which is best and https://docs.fluentd.org/filter#filter-chain-optimization says to not use filter_stream ... is there any reason to keep it or can it be changed as long as the tests pass ?

I defer to you here. I know of no reason it needs to be implemented as-is. Refactoring is acceptible IMO:

• if it makes the code more readable
• tests continue to pass
• It meets recommended plugin guidelines

[cosmo0920]

From this PR's benchmarks, using filter chain optimization makes around 10 ~ 20% faster than using #filter_stream.

But I have no idea that using #filter method whether improves code readability and continue to pass unit tests or not.

[grosser]

I started using https://github.com/grosser/fluent-plugin-kubelet_metadata (which also uses filter)so no longer working on this, can close if nobody else wants it.